Lift Off Det' - causes and cures

Hi Andre, I think we are experiencing lift off det. Simtek ECU flashes CEL when lift off between gear changes at high load/rpm and occasionally drops into a ‘limp mode’ (though in actual fact the only things in software which can induce limp mode are low fuel pressure or EOP and the ECU has neither sensors connected). Not familiar enough with internal logging on this ECU to know what channels to log to know if knock sensor raw signal is exceeding knock threshold settings (would make sense if there was simply a binary ‘knock detected’ channel).

What is the primary cause of ‘lift off det’? And how should we tackle it?

Engine is EJ20. Boost is 1 bar. Fuel mixtures are about 0.8 on full boost. Ignition timing is conservative. Timing at high rpm, light load is set to a block of 36°. Would reducing this help?

To be honest, lift off detonation isn't something I've ever experienced and when you think about the conditions present, there's really nothing that can cause it - You've got a closed throttle and low cylinder pressure. Occasionally I come across an engine that exhibits tip-in knock which I'm wondering if you're confusing with the same thing? This is where the engine knocks briefly when you get back on the throttle after a gear change. It's very rare in my experience and I've had perhaps a handful of cars that have done this. Some ECUs (I Can't speak for simtek) have a tip-in retard function specifically to address this. Out of interest how are you monitoring knock?

Your limp home problem is a little confusing and I think you need to carefully log any cuts as well as the injector pulse width and ignition timing around the points this is happening as this will give you more information to work with.

Bit of an update. The ECU reads a knock raw value, there is a threshold which triggers CEL, ignition retard and fuel enrichment and then progressively tries to reinstate normal timing and fuelling. There is also a higher knock threshold 'no restore' which if triggered will maintain the 5 degrees retard and fuel enrichment (as well as it seems pulling all the duty from the WG) until the ECU is power cycled. This is not considered part of the limp mode, just part of the knock strategy.

The ECU does per cylinder knock detection with 'windowing' and is set to something like 10 degrees before to 30 degrees after TDC, it is possible to turn off individual cylinders in this respect.

After talking to Steve at Simtek, he says that he has experienced events being picked up as 'phantom' knock on lift off which he puts down to mechanical noise on engines which are built on the loose side, where a cylinder other than the one the ECU is 'listening for' has some slap at BDC? To be honest, I don't think that is the case on this car as it's a bone stock engine and I couldn't hear anything untoward through the det cans accept for a few 'clicks' at idle which sounded to me exactly like mild knock (very odd) but sounded clean enough through the revs on boost.

For knock detection while tuning I use a Phormula KS-Pro in conjunction with a KS-4. I tune mostly Subaru and I've got in the habit of removing the OE sensor and putting the Phomula in its place while tuning as I've found few alternative positions on the block. Once I'm happy the engine is tuned without knock, I'll put the OE sensor back. Whether this is the best strategy I don't know. Traditionally gravel rally cars we'll leave knock detection turned off anyway due to the theory that stones hitting the sump gaurd can falsely trigger the knock detection.

The next step is to log the knock raw and other knock parameters as well as obviously throttle, boost, revs and advance on the on-board logging. We do now have a wideband connected to the ECU too but at the time of tuning were using just an LM2.

I'm tempted also as a sanity check to get the timing light out too. It's coil on plug and previous experience tells me it's easier to put an ignition lead between the plug and COP coil to get the trigger for the inductive clamp.

It is common to see large knock sensor activity when you lift off from high RPM, I have seen this with many different platforms. Obviously it is not knock as you will have very low MAP and almost no combustion when you snap the throttle closed, so most likely mechanical noise from somewhere. I always assumed it was backlash in the clutch spline/multiplates or trans as I havent seen it with an auto trans car yet but I guess it could be something else I hadnt considered such as piston slap or timing chain etc. Most ECUs have some type of "lockout" to disable the knock control during decel.

Thanks for your input Adam. There is in Simtek a TPS threshold for knock detection so I'll check what that is set to and maybe raise the threshold so knock detection is only 'on' at higher throttle openings - might be safer than rising the overall knock threshold.